|
| |||||||||||||||||||||||||||||||||||||||
Background The Björnstad syndrome, an autosomal recessive disorder associated with sensorineural hearing loss and pili torti, is caused by mutation of a previously unidentified gene on chromosome 2q34–36.
Methods Refined genetic mapping and DNA sequencing of 44 genes between D2S2210 and D2S2244 revealed BCS1L mutations. Functional analyses elucidated how BCS1L mutations cause the Björnstad syndrome.
Results BCS1L encodes a member of the AAA family of ATPases that is necessary for the assembly of complex III in the mitochondria. In addition to the Björnstad syndrome, BCS1L mutations cause complex III deficiency and the GRACILE syndrome, which in neonates are lethal conditions that have multisystem and neurologic manifestations typifying severe mitochondrial disorders. Patients with the Björnstad syndrome have mutations that alter residues involved in protein–protein interactions, whereas mutations in patients with complex III deficiency alter ATP-binding residues, as deduced from the crystal structure of a related AAA-family ATPase. Biochemical studies provided evidence to support this model: complex III deficiency mutations prevented ATP-dependent assembly of BCS1L-associated complexes. All mutant BCS1L proteins disrupted the assembly of complex III, reduced the activity of the mitochondrial electron-transport chain, and increased the production of reactive oxygen species. However, only mutations associated with complex III deficiency increased mitochondrial content, which further increased the production of reactive oxygen species.
Conclusions BCS1L mutations cause disease phenotypes ranging from highly restricted pili torti and sensorineural hearing loss (the Björnstad syndrome) to profound multisystem organ failure (complex III deficiency and the GRACILE syndrome). All BCS1L mutations disrupted the assembly of mitochondrial respirasomes (the basic unit for respiration in human mitochondria), but the clinical expression of the mutations was correlated with the production of reactive oxygen species. Mutations that cause the Björnstad syndrome illustrate the exquisite sensitivity of ear and hair tissues to mitochondrial function, particularly to the production of reactive oxygen species.
BCS1L, a 419-amino-acid chaperone protein, is a member of the family called AAA, an acronym for ATPases associated with various cellular activities.5,6 The AAA-family ATPases mediate the folding, unfolding, assembly, and degradation of proteins.7,8 Found within the inner mitochondrial membrane, BCS1L is presumed to facilitate insertion of Rieske Fe/S protein into precursors to complex III5 during assembly of the respiratory chain. Complex III then becomes assembled with complexes IV and I into a respirasome supercomplex that facilitates the electron transfer required for the synthesis of ATP.9,10
BCS1L mutations have previously been reported to cause complex III deficiency1 (manifested by neonatal tubulopathy, encephalopathy, and liver failure; Online Mendelian Inheritance in Man [OMIM] number 606104) and the more severe disorder of intrauterine growth retardation, aminoaciduria, cholestasis, iron overload, lactic acidosis, and early death, termed the GRACILE syndrome (OMIM number 603358).11,12 The clinical findings of these profoundly severe multisystem disorders are more typical of gene mutations that disrupt the mitochondrial respiratory-chain complex than are the highly restricted clinical findings associated with the Björnstad syndrome.
To understand how BCS1L mutations cause such widely different clinical phenotypes, we considered the locations of defects on the BCS1L protein structure and compared the function of mutant BCS1L in yeast and in human lymphocytes. These studies showed that BCS1L mutations altered assembly of the mitochondrial respirasome, reduced activity of the electron transport chain, and increased the production of reactive oxygen species. The production of reactive oxygen species was correlated with the clinical severity of different BCS1L mutations. These data indicate that in addition to mitochondrial heteroplasmy and variable energy requirements of tissues, tissue-specific sensitivities to reactive oxygen species contribute to the variability of the manifestations of mitochondrial defects.
Methods
DNA Sequencing and Mutation Analysis
We amplified genomic DNA, prepared as previously described,4 using primers flanking exon–intron gene boundaries, and we sequenced the DNA using the ABI BigDye method in an Applied Biosystems Prism 3700 DNA Analyzer. Sequence variants were confirmed by restriction-enzyme digestion, and polymorphisms were identified in National Institutes of Health (NIH)13 and Celera14 databases.
Immunohistochemistry
Paraffin-embedded sections of human scalp and mouse inner-ear tissue were prepared for antigen retrieval as previously described.15 Scalp sections were reacted with antibodies to BCS1L (Proteinech Group) and to core protein 1 (Molecular Probes), followed by antigen detection with a fluorescein-conjugated biotin–streptavidin system (Vector Laboratories). Binding of antibody to murine inner-ear sections was detected with alkaline phosphatase and NBT/BCIP substrate (Roche). Fluorescent images were captured with a confocal microscope (Leica Microsystems) and processed with NIH Image and Adobe Photoshop software.
Complementation Studies with Yeast
Mutations were introduced into human BCS1L by means of a site-directed mutagenesis kit (QuikChange, Stratagene) that used overlapping primers containing the desired mutation that were introduced into the pMGL3 vector.16 Constructs were transfected into Bcs1-deficient yeast (aW303
Bcs1) and grown on rich (YPDextrose) or mitochondria-selective (YPEthanol/Glycerol) media at 30°C for 4 days. The size and growth of the colony (measured by log-phase doubling times of cultures; initial concentration, 1x105 cells per well) were compared between wild-type (W303-1A) yeast and aW303
Bcs1 transfected with wild-type yeast Bcs1, wild-type human BCS1L, empty vector (pMGL3), or human mutant BCS1L.
Mitochondrial Extraction, Analysis, and Content
Actively growing transformed lymphocytes were washed, and a crude mitochondrial fraction was extracted as previously described.17 Mitochondrial preparations, isolated in the presence or absence of magnesium plus either ATP or
S-ATP, were solubilized in 0.75 M
-aminocaproic acid, 50 mM Bis-Tris (pH 7.0), and 5 mg of digitonin per milliliter (all from Sigma). The complexes underwent blue-native polyacrylamide-gel electrophoresis as previously described.9 Western blots of protein complexes were probed with antibodies to BCS1L, core protein 1, and Rieske Fe/S proteins. Before undergoing Western blot analyses with Rieske Fe/S protein antibody (Invitrogen Molecular Probes), samples were run on a second dimension of sodium dodecyl sulfate–polyacrylamide-gel electrophoresis. Mitochondrial content was assessed by staining transformed lymphocytes with MitoTracker Green FM and hydroethidium to normalize for cell number (Invitrogen Molecular Probes).
Electron-Transport-Chain Activity
The rate of reduction of resazurin dye, which was dependent on the electron-transport chain, was measured by spectrophotometry of lymphocytes treated with the F1-ATP synthase inhibitor oligomycin as previously described.18 The mean (±SE) values from at least three assays are reported.
Production of Reactive Oxygen Species
Production of reactive oxygen species was assessed in isolated mitochondria treated with complex-specific substrates, with inhibitors and without inhibitors, and analyzed with the use of the hydrogen peroxide–sensitive dye Amplex Red (Invitrogen Molecular Probes).19 The substrates were glutamate and malate for complex I and succinate for complex III. The inhibitors were rotenone for complex I and antimycin A plus rotenone for complex III (Sigma-Aldrich). Superoxide was monitored as hydrogen peroxide after the addition of superoxide dismutase (Sigma-Aldrich) to mitochondrial isolates. The values were controlled for mitochondrial content and normalized to cell number. The mean values from at least three assays are reported.
Results
Gene Sequencing
We previously identified the locus of the autosomal recessive gene for the Björnstad syndrome1,2 on chromosome 2q34–36.4 Fine mapping in a consanguineous family with the Björnstad syndrome (BS1) (Figure 1A) narrowed the critical disease region to a 2-megabase region between D2S2210 and D2S2244 that contains 47 genes (Figure 1B). DNA sequences encoding 44 genes were determined in BS1 family members. After previously reported polymorphisms had been excluded,13,14 one sequence variant remained in BCS1L. The single-nucleotide variant in BCS1L substituted a histidine for an arginine (R183H) and was absent in 300 normal chromosomes.
|
|
We also identified two BCS1L mutations (G35R and R184C, designated CIII) in a 4-year-old child who had clinical manifestations of the Björnstad syndrome as well as growth retardation, developmental delay, and profound hypotonia. These multiorgan phenotypes resembled those of mild complex III deficiency.1,11,12 Spectrophotometric analysis of enzymatic activities of electron-transport complexes21,22,23 from skeletal muscle showed that complex III activity was reduced to 15% of normal. Hearing loss, but not pili torti, occurs in complex III deficiency12; neither is a typical feature of the GRACILE syndrome, perhaps because recognition of the syndrome is hindered by complex systemic manifestations and early death.
We considered how the type and location of BCS1L mutations contributed to the different human phenotypes. BCS1L has two functional domains (Figure 2B): a sorting sequence (amino acids 1 through 89) that facilitates translocation and insertion into the mitochondrial inner membrane,24,25 and the domain of the AAA-family ATPase (residues 220 through 399).
Two mutations associated with the Björnstad syndrome and one mutation associated with complex III deficiency and the GRACILE syndrome encoded truncated BCS1L peptides; these mutations are presumed to be functionally null alleles (Figure 2B). Two other reported mutations associated with complex III deficiency and the GRACILE syndrome, as well as the CIII mutation, affected sorting sequences; by preventing BCS1L from reaching the mitochondrial inner membrane, these mutations may also be functionally null alleles. Each patient with the Björnstad syndrome carried at least one allele that was predicted to have partial function.
We located five BCS1L mutations within the AAA domain on the three-dimensional structure (Figure 2D) of Escherichia coli RuvB,20 which has the highest sequence homology to BCS1L of all AAA domains.26 Three mutations associated with complex III deficiency and the GRACILE syndrome, including a missense mutation (S277N) in the Walker B domain,27 altered residues that directly interact with magnesium or ATP, a finding that implied that these defects impair nucleotide binding or hydrolysis. In contrast, mutations at residues 302 and 306 in patients with the Björnstad syndrome altered an exterior face of the AAA domain, a location that suggested that these defects affect interactions between BCS1L and other proteins.
Immunohistochemical Analysis
We examined the expression of complex III by means of immunohistochemical analysis using antibodies specific to BCS1L and core protein 1, a component of complex III. Concentric layers of mitochondria-rich cells with abundant amounts of BCS1L and core protein 1 were most prominent at the dermal papilla (Figure 3A) but also surrounded the hair shaft throughout the length of the follicle in sections of skin from the normal human scalp. In the mouse inner ear, antibodies to BCS1L and core protein 1 produced strong immunostaining in the hair cells and the stria vascularis (Figure 3B). Although the tissues in which these proteins were expressed are those that are involved in the Björnstad syndrome, BCS1L was also highly expressed in heart and brain tissues (data not shown).
|
Because of the location of the mutations associated with the Björnstad syndrome and the ATPase activity of BCS1L, we examined whether mutations affected ATP-dependent protein–protein interactions. BCS1L-associated complexes were extracted with ATP or with
S-ATP, a nonhydrolyzable ATP analog, and analyzed by electrophoresis and Western blotting. In normal mitochondria, BCS1L antibody identified three complexes (Figure 4A). The smallest species had a molecular weight of approximately 240 kD, a size consistent with a hexameric ring of BCS1L monomers (40 kD) (Figure 4C), the multimeric structure recognized in other AAA-family ATPases.28 No BCS1L mutations disrupted this structure.
|
S-ATP, BCS1L antibody also labeled two complexes with higher molecular weights that were designated BCS1LH1 (approximately 400 kD) and BCS1LH2 (approximately 450 kD). Because both species were missing from extracts of mitochondria with the CIII mutation, these mutations prevented the formation of protein species that required ATP binding and hydrolysis (Figure 4C). Mitochondrial extracts from patients with the Björnstad syndrome had abnormal ratios of BCS1LH1 and BCS1LH2 (Figure 4A). Because ATP regulates the chaperone activity of AAA-family ATPases28 and promotes dissociation of BCS1L from complex III,5 the mutations associated with the Björnstad syndrome may alter the ATP-dependent equilibrium between BCS1L within a hexameric ring or in BCS1LH1 and BCS1LH2 (Figure 4C). The consequence of these altered levels was the diminished assembly of complex III and the accumulation of unassembled Rieske Fe/S protein.
To examine whether different BCS1L mutations affected assembly of the respirasome, a molecular supercomplex, mitochondrial complexes were fractionated by electrophoresis and probed with antibodies (Figure 3B). Although all extracts contained a fully formed respirasome — (CIII)2(CIV)n(CI)n — the levels and mobility of intermediate species of respirasome differed among extracts from normal mitochondria, mitochondria from patients with the Björnstad syndrome, and mitochondria from patients with the CIII mutation. Normal mitochondrial extracts had prominent complex III dimers that contained Rieske Fe/S protein (denoted [CIII]2). In contrast, smaller complex III species without Rieske Fe/S protein (denoted [apoCIII]2) accumulated in mitochondrial extracts from BCS1L mutant lymphocytes. Intermediate respirasome complexes containing complex III and complex IV multimers were most prominent in extracts from CIII lymphocytes; lesser amounts were also found in extracts from patients with the Björnstad syndrome. No intermediate supercomplex species from mutant extracts were labeled by antibody to Rieske Fe/S protein (data not shown), a result indicating that mutant extracts accumulated (apoCIII)2(CIV)n intermediates without Rieske Fe/S protein. This conclusion was supported by the abundant accumulation of unassembled Rieske Fe/S protein in mitochondrial extracts from lymphocytes of patients with the Björnstad syndrome or with the CIII mutation, but not in extracts from normal controls (Figure 4B, bottom).
Complementation Studies with Yeast
To assess the functional consequences of human BCS1L mutations, we performed yeast complementation studies. Yeast lacking Bcs1, the BCS1L homologue,5 fail to incorporate Rieske Fe/S protein into complex III and do not grow on media that require respiratory-chain metabolism.6 This deficiency is complemented by normal human BCS1L29 but not by BCS1L from patients with mutations associated with complex III deficiency or the GRACILE syndrome.12,30 We examined the growth of Bcs1-deficient yeast transfected with mutations causing the Björnstad syndrome and other BCS1L mutations (Figure 5A). Normal BCS1L supported yeast growth, but BCS1L with Björnstad syndrome mutations, the CIII mutation, and mutations associated with other complex III deficiencies attenuated growth on selective media to about the same degree. Although these data confirmed that mutations associated with the Björnstad syndrome cause mitochondrial dysfunction, they did not suggest why different mutations resulted in distinct clinical phenotypes.
|
We examined the functional consequences of BCS1L mutations for electron-transport-chain activity in lymphocyte mitochondria (Figure 5B). Normal electron-transport-chain activity was reduced in mitochondria from patients with the Björnstad syndrome (to 62% of normal, P=0.07) and in mitochondria with the CIII mutation (to 50% of normal, P=0.01). Reduced electron-transport-chain activity did not reflect mitochondrial content. Lymphocytes with the CIII mutation had 16% more mitochondria per cell than lymphocytes from patients with the Björnstad syndrome or normal lymphocytes (P=0.02).
Production of Reactive Oxygen Species
We assessed the production of reactive oxygen species from complexes I and III.31 In comparison with the amount of superoxide produced by normal lymphocyte mitochondria, the amount of superoxide produced by complex III activity was reduced by 61% in mitochondria from patients with the Björnstad syndrome (P=0.01) and by 68% in mitochondria from patients with the CIII mutation (P=0.02); the amounts of these reductions paralleled diminished electron-transport-chain activities. Complex I produces negligible superoxide in the normal respirasome,32 but complex I in mitochondria with mutations causing the Björnstad syndrome produced 30 times as much superoxide as normal (P=0.01) (Figure 5C). The amount of superoxide produced by complex I in mitochondria with the CIII mutation was still higher (50 times as much as normal, P=0.01) and was significantly greater than that produced by mitochondria with mutations causing the Björnstad syndrome (P=0.04).
Discussion
We have shown that recessive BCS1L mutations cause sensorineural hearing loss and pili torti in the Björnstad syndrome. Biochemical and functional studies of mutant BCS1L showed that this chaperone protein has essential roles in the assembly of complex III and the respirasome. Mutations associated with the Björnstad syndrome and CIII mutations in BCS1L resulted in the accumulation of mitochondrial respiratory intermediates that lacked Rieske Fe/S protein (Figure 4) and decreased electron-transport-chain activity. CIII mutations also increased mitochondrial content. Although all BCS1L mutations increased the production of reactive oxygen species, the amount of the increase was specific to the mutation and the disease.
The severity of the medical manifestations of BCS1L mutations was inversely correlated with the amounts of the high-molecular-weight mitochondrial complexes BCS1LH1 and BCS1LH2. These ATP-dependent complexes were missing in mitochondria with the CIII mutation, which confirmed structural predictions that these clinically severe BCS1L mutations impaired ATP binding or hydrolysis. In contrast, mutations causing the Björnstad syndrome only altered the normal ratios of high-molecular-weight BCS1L complexes, a finding consistent with the location of these mutations in residues on the external chaperone surface where protein–protein interactions may occur. Although we could not identify core protein 1, a component of complex III, in either BCS1LH1 or BCS1LH2, we hypothesize that these species may represent precursors of BCS1L or complex III, BCS1L hexamer conformations, or BCS1L bound to unidentified chaperones of complex III assembly. Nonetheless, the consequence of all BCS1L mutations was accumulation of complex III precursors that lacked the Rieske Fe/S protein. Remarkably, the assembly of respirasomes continued despite diminished chaperone activities of BCS1L, albeit with the accumulation of atypical intermediates such as (apoCIII)2(CIV)n.
Because unassembled respirasome complexes and free Rieske Fe/S protein occurred with all BCS1L mutations, these are unlikely to be major determinants of clinically disparate diseases. Furthermore, electron-transport-chain activity, as assessed indirectly by complementation analysis in yeast or in mitochondria from mutant lymphocytes, differed only moderately between mitochondria with mutations causing the Björnstad syndrome and mitochondria with CIII mutations. However, only CIII mutations led to a compensatory increase in the mitochondrial content of lymphocytes, which probably reflects a more severe electron-transport-chain defect.
An untoward effect of inadequate electron-transport-chain activity and increased mitochondrial content was increased production of reactive oxygen species. We found that all BCS1L mutations reduced complex III–mediated production of superoxide, which reflected diminished assembly of complex III due to impaired insertion of Rieske Fe/S protein; this model is supported by studies that directly implicate Rieske Fe/S protein in the production of reactive oxygen species.33 In contrast, production of superoxide by complex I was strikingly higher in mitochondria with CIII mutations than in those with mutations associated with Björnstad syndrome (P=0.04) (Figure 5). Whether increased production of superoxide by complex I reflected increased electron-transport-chain activity upstream of complex III (potentially further augmented by increased mitochondrial content in CIII mutations), toxic effects of abnormal respirasomes, or changes in chaperone activity, the degree to which BCS1L mutations increased the production of reactive oxygen species predicted disease phenotype.
We conclude that the BCS1L mutations that affect ATP binding or hydrolysis and severely disrupt complex III assembly cause disease by attenuating electron-transport-chain activity and markedly increasing the production of reactive oxygen species. Despite increased mitochondrial content, these mutations result in the multisystem organ failure that characterizes complex III deficiency and the GRACILE syndrome.
We suggest that the BCS1L mutations responsible for the highly restricted phenotype of the Björnstad syndrome increase the production of reactive oxygen species by complex I, which leads to oxidative stress in both the inner ear and the hair follicle. Current models of the effects of aminoglycoside antibiotics and excessive noise on hearing support a critical role for increased levels of reactive oxygen species in ototoxicity.34 We presume that hair-follicle cells that divide rapidly may also be particularly susceptible to increased levels of reactive oxygen species or to other respiratory-chain defects in patients with the Björnstad syndrome. We also note that mitochondrial disease is often manifested by a variety of hair abnormalities.18 Furthermore, we speculate that the remarkable tissue-specific manifestations of mutations leading to the Björnstad syndrome may hint at a shared mechanism for age-related hair abnormalities and hearing loss, since aging, like the BCS1L mutations associated with the Björnstad syndrome, increases the levels of reactive oxygen species.31
Supported by grants from the Howard Hughes Medical Institute and the National Institutes of Health.
No potential conflict of interest relevant to this article was reported.
This work is dedicated to the memory of our colleague Alfonso Esparza.
We thank the patients and family members for their participation, M.A. Flores and R.M. Caldera for clinical assistance, Alexander Tzagoloff for providing plasmids and yeast strains, and Fred Winston and Lisa Laprade for technical assistance.
Source Information
From Harvard Medical School, Boston (J.T.H., V.R.F., J.S., B.M., I.K., A.E., Y.N., R.D.E., J.G.S., C.E.S.); the Howard Hughes Medical Institute, Boston (J.T.H., J.S., J.G.S., C.E.S.); University Hospital Würzburg and Institute of Clinical Biochemistry and Pathobiochemistry — both in Würzburg, Germany (J.S.); Aalesund Hospital, Norway (N.B., G.S.); Hammersmith Hospitals and Imperial College, London (P.S.); the Massachusetts Eye and Ear Infirmary, Boston (I.K., A.E., Y.N., R.D.E.); Instituto de Ciências Biológias e da Saúde, Pontifícia Universidade Cathólica de Minas Gerais, Belo Horizonte, Brazil (R.G.); University of Washington, Seattle (F.S.); H:S Bispebjerg Hospital, Copenhagen (E.S., L.T.); the Brain Tumor Institute, Cleveland Clinic, Cleveland (B.H.C.); Louis Stokes Veterans Affairs Medical Center, Cleveland (C.L.H.); and Institute of Medical Biochemistry and Genetics, University of Copenhagen, Copenhagen, and University Hospital of Tromsø, Tromsø, Norway (L.T.).
Drs. J.G. Seidman and C.E. Seidman contributed equally to this article.
Address reprint requests to Dr. C.E. Seidman at the Department of Genetics, Harvard Medical School, Room 256 NRB, 77 Ave. Louis Pasteur, Boston, MA 02115, or at cseidman{at}genetics.med.harvard.edu.
References
| |||||||||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |